Design And Implementation Of Eye Tracking Syetem

Eye tracking methodology is not only the most effective way of visual information processing, but also one of the key methodologies of future intelligent human-machine interface. There is a broad application in industry control, robotics and clinical medicine. Eye tracking methodology based on pupil-cornea tracking principle and image processing gets the advantage of accurate measurement.The work of this paper includes principle and method introduction about eye tracking methodology, structure of eye tracking system, hardware design and software design based on VC++6.0 about eye tracking system.For data acquisition from dynamic eye movement image, a new algorithm named center acquisition from pupil and corneal reflex is proposed in this paper according to pupil-corneal tracking principle.First, under the infrared light, the eye image was captured by camera. In order to reduce the processing time, the image processing area was acquired by applying adaptive threshold. Second, the center of corneal reflex was extracted by using a high threshold value and a low threshold value.Then, the optimum adaptive threshold value was calculated to get the location and size of pupil. Finally, the feature points of the edge of pupil were got by applying gradient method and the center of the pupil is located by fitting ellipse. The results show that this algorithm could not only guarantee the precision and stability of getting the center of pupil and the center of croneal reflection but also met the demand of real-time processing.For demarcating fixation, nonlinear second order polynomial function fitting method is applied to get the one-one mapping between eye image coordinate system and scene image coordinate system.Fixation will be calculated though mapping function and coordinate of pupil-cornmeal. Then save and display fixation in scene image. It is indicated that the experiment result is with a high precision and the average distance error is 8.7 pixels.In order to analyze the distribution of the fixation, the order and the frequency that an eye looks at the objective, a gaze trajectories is drawn by VC program in which Matlab components are called to realize the hot pot map and geomorphologic map. In that case, the eye movement will be showed more clearly and effectively.